一种特征约束的有限元人脑图像配准方法

提出了一种结合图像的解剖标记点和自适应有限元网格进行人脑图像的精确配准方法.首先利用Forstner算子提取对应图像的解剖标记点,并作初始的图像刚性变换.为了使有限元网格能更加准确地刻画图像解剖结构分布特征,本文利用图像的梯度分布建立了自适应的有限元网格剖分,结合标记点作为有限元的形变约束,使得配准的精度和有限元的计算效率得到提高.人脑图像配准的实验结果表明,该方法能有效地解决图像弹性配准问题.

Finite Element Approach Based on Feature Constraint for Elastic Image Registration of Human Brain

A hybrid registration method based on adaptive finite element (FEM) deformation and anatomical landmarks is presented. Anatomical landmarks are extracted by the image geometry feature operator introduced by Forstner. These landmarks determine the rigid transformations between corresponding images before FEM is used for the restrictions of the FEM deformation. The adaptive finite mesh is generated according to the gradient field of brain atlas, so the mesh can also represent structure features in the atlas. Result shows that the registration is more accurate under the same conditions.